Antarctic snowfall is the main mass input to the Antarctic Ice Sheet, and its estimate relies mainly on satellite remote sensing. Unfortunately, because of ground clutter, the two W-band Cloud Profiling Radars (CPRs) on board CloudSat (NASA, 2006–2023) and EarthCARE (ESA–JAXA, 2024–ongoing) cannot reliably observe the lowest ~1000 m and ~500 m of atmosphere respectively ("blind zone"). Surface snowfall must then be estimated by assuming that what is detected aloft reaches the ground. But in katabatic-driven coastal East Antarctica this assumption is frequently violated by virga, precipitation that sublimates completely before reaching the surface. We analyse four years (2020–2021, 2023–2024) of vertical Doppler radar profiles (MRR-2, 24 GHz) and co-located optical disdrometer data at the Italian Mario Zucchelli Station. We identify virga profiles when hydrometeors detected aloft produce no signal at the lowest radar gate (105 m) or in the disdrometer. Virga accounts for 38 % of the hydrometeor-bearing columns. At the EarthCARE-like lowest observable heights, part of the virga becomes indistinguishable from precipitation and the four-year surface accumulation is overestimated by 29–30 % (up to 34 % switching to CloudSat). ERA5 reproduces the dry layer but still deposits snowfall in 90 % of the virga hours, accumulating on average 156 mm yr⁻¹ against 53–106 mm yr⁻¹ from the radar-based estimates. These results quantify the blind-zone impact at a coastal Antarctic site, sampling unprecedentedly close to the ground, and establish the ground-based classification, released as an open-source package, as a transferable constraint for satellite and reanalysis products.